The Discharge Gap: why England’s stroke care stops at six months and the four actions that would fix it

England built one of the world’s best acute stroke pathways. Then, at six months, it stops, stops treating, stops measuring, and stops being accountable. The Discharge Gap, an independent policy report I have co-written with fellow survivor and patient leader Maya Kuzalti, sets out why that cliff exists, what it costs, and the four actions that would close it.

The report is published today in full below: the original designed edition, a downloadable PDF, and a short, accessible version written for stroke survivors and carers. It was written without commission, sponsor or funder, and carries a foreword by Professor Nick Ward, Professor of Clinical Neurology at the UCL Queen Square Institute of Neurology. Its scope is England: health is devolved, and the funding and commissioning arguments do not transfer automatically to Scotland, Wales or Northern Ireland.

Read the report

Open The Discharge Gap full screen on Canva · Maya Kuzalti & Paul Quinn, October 2026

The cliff, in numbers

A stroke happens in the UK approximately every five minutes,  around 100,000 a year, and 1.4 million people are living with the consequences. The acute response is genuinely world-class: the median time from arrival to thrombolysis in 2024/25 was 56 minutes (SSNAP, 2025). What follows is not.

• 35.1% of eligible survivors received the recommended six-month review in 2024/25, down from 38.8% the year before and falling since 2019/20 (SSNAP, 2025).
• Physiotherapy is delivered on 70% of recommended days in hospital, 15% in community rehabilitation, and in life after stroke there is no national target at all (CSP/ACPIN, 2026).
• Even inpatient therapy ran far below guideline: a median of 13.8 minutes of physiotherapy, 12.9 of occupational therapy, 6.7 of speech and language therapy and 1.9 of psychology per inpatient day, against 45 minutes of each (Gittins et al., 2020, SSNAP data 2013–15). The current guideline asks for at least three hours a day.
• Community stroke services run with 26% fewer registered physiotherapists and 36% fewer physiotherapy support workers than guidance requires (CSP/ACPIN survey of 159 services).

After six months, SSNAP, the national stroke audit, stops counting. There is no national data point at twelve months, two years or five. In a data sense, a survivor who leaves the measurement window has recovered. The system has simply stopped asking.

Three decisions built the cliff

The report’s central finding is structural, not a funding mystery. Three decisions, each defensible on its own terms and taken in different rooms, combine to create a gap that no single organisation owns:

1. Measurement ends at six months. SSNAP was built in 2012 around acute care. The six-month follow-up was a later extension; nothing sits beyond it.
2. The right guidance has no teeth. NICE NG236 (2023) and the 2023 National Clinical Guideline explicitly reject the six-month “plateau” and say survivors can benefit from rehabilitation at any point. But NICE guidance is advisory, not a funding mandate.
3. Commissioning has not followed. Many ICBs do not commission community rehabilitation against NG236, because nothing in their contracts, payment mechanisms or performance measures requires them to, and without data beyond six months, nobody can hold them to it.

The payment architecture compounds the problem. Acute care is paid on tariff with Best Practice Tariff incentives; GP practices through QOF. Community health services, where stroke rehabilitation is actually delivered,  sit almost entirely on block contracts, paid a fixed sum irrespective of activity or quality, and NHS England’s own 2025 baseline exercise found non-acute data “very limited” (HFMA, 2026). Commissioners cannot see community activity, let alone outcomes.

Then there is the double standard. New drugs are appraised against a published £25,000–£35,000 per QALY threshold and must be funded within 90 days of a positive NICE recommendation; they do not have to save the NHS money to pass. Rehabilitation has no equivalent national route: it competes for cash-limited local budgets with no national threshold to be appraised against and no deadline by which a positive recommendation must be funded. As Professor Ward puts it: “So why is rehabilitation held to a different account?”

Where the money actually sits

The Stroke Association’s 2026 model, built by Lane Clark & Peacock, puts the UK cost of stroke at £25.71 billion in 2025, rising to £35.86 billion by 2035 in 2025 prices. The composition is the point:

Component (2025, £bn)	Cost	Share
Social care	13.72	53%
Unpaid production and informal care	5.69	22%
NHS healthcare	5.34	21%
Lost paid productivity	0.95	4%
Total	25.71	100%

About 79% of the modelled cost sits outside the NHS healthcare budget that would pay for better rehabilitation, and 87% of NHS and social care cost arises after the first year. This is a classic misaligned-incentive problem: the commissioner who funds therapy carries the cost, while the savings, fewer care packages, fewer carer breakdowns, land with councils and households.

The report is explicit about the limits of the figures. They are modelled estimates, not official statistics; the Stroke Association itself calls £25.7 billion “the lowest defensible estimate”. The model prices current provision rather than need, so it cannot show what the six-month cliff costs or what better rehabilitation would save. That is precisely why the report proposes a bounded pilot that measures those costs rather than claiming a saving.

The evidence already exists

Each piece of evidence is presented with what it shows and, just as carefully, what it does not:

• Late recovery is real. Queen Square’s upper limb programme delivered 90 hours over three weeks to people a median 18 months after stroke; by six months, 68.3% exceeded the minimum clinically important difference on the Fugl-Meyer scale (Ward et al., 2019). Uncontrolled, single-centre, and a younger cohort — a randomised trial is now under way — but a strong case for testing higher doses.
• Cognition is being missed. Imperial’s IC3 battery found impairment in survivors scoring “unimpaired” on the MoCA, at a mean rate of 30.2%, and tracked everyday function more closely (Gruia et al., 2025). Evidence on detection, not treatment — and not adopted anywhere in the NHS pathway.
• Spasticity is common and undertreated. 45.7% of survivors with paresis had increased tone within three months (EPITOME interim data, Ipsen-sponsored); prevalence beyond six months is 24.2% (Zeng et al., 2021). There is no England treatment figure at all — itself a finding.
• A null result, honestly reported. The RETAKE trial of early vocational rehabilitation found no significant difference in return to work at 12 months (64.2% v 59.4%). It tested early intervention, not rehabilitation beyond six months, and the report cites it as a design lesson, not a verdict.
• International mechanisms. Ontario offers a named coordinator and re-entry up to a year after discharge; Sweden’s Riksstroke register follows survivors at three and twelve months; Singapore has piloted a coordinator model. None proves causation; together they show the components England lacks are workable.

What 20 survivors told us

An indicative (non-probability) survey of 20 survivors found 12 received no discharge or care plan; 14 described post-discharge support as little or nothing; 18 turned to charities, private providers or personal networks; and 9 of 19 paid out of pocket or went without. Asked what would have helped, 12 wanted longer rehabilitation, 11 a named clinical contact and 10 a scheduled review — an indicative design specification for the pathway the report proposes.

“My memory of it is that basically when you discharge, you just get pushed off a cliff and left to get on with things.”

Stephen Rimmer, stroke survivor

Four actions, four owners

1. HQIP and NHS England: extend SSNAP to twelve months. The cheapest step. SSNAP runs on a contract worth about £760,000 a year, and the tender reserved up to £1m for changes of scope, so a twelve-month data point could be a contract variation rather than a new programme — subject to the extension terms, which HQIP would need to confirm. Then attach a consequence: a twelve-month metric in the NHS Oversight Framework, or re-pointing part of the Best Practice Tariff towards outcomes.
2. NICE: publish auditable standards for the community components of NG236, with explicit re-entry criteria and an outcome measure better suited to rehabilitation than the modified Rankin Scale alone.
3. ICBs with local authorities: commission and pilot. A bounded pilot in two or three ICBs, funded by redirecting existing community stroke spend and pooled under Section 75 or the Better Care Fund so social care and carer savings count in the same business case. A named coordinator, GP referral back in for up to twelve months, carers identified and assessed. If it cannot show a twelve-month benefit or lower caregiver strain, it stops.
4. Survivors and carers as co-designers — in paid roles with decision-making authority, seated before the commissioning model is set, not consulted once it is fixed.

None of this requires a new national budget line. As the report concludes: the case is not that this is affordable. The case is that the current arrangement is not.

---

The Discharge Gap is an independent publication by Maya Kuzalti and Paul Quinn, October 2026, written without commission, sponsor or industry funding. The authors hold voluntary roles as Lay Research Stakeholders at Imperial College Healthcare NHS Trust and Ambassadors for UK Stroke Charities; none of these organisations reviewed or endorsed the report. Full methodology, disclosures and references are in the report. Follow the project at thedischargegap.substack.com or write to hello@thedischargegap.com.